Why’s Everyone So Bearish, Robinhood's AI Agent and SoFi's Stablecoin Launch

Summary

• This bear market is better than 2018’s on fundamentals but worse psychologically, and the hosts split hard on why. Santi recalls that in 2018 he and Mike debated getting jobs at Uber or Netflix if the industry died; today the question is only “what will we do if the tokens don’t go up again” — whether value accrues to tokens or to equity, not whether public blockchains work. Santi’s darker counter: “what’s worse than being dead is being irrelevant,” arguing after reviewing ~30 sectors that stablecoin use cases are far narrower than the industry assumed.
• Santi’s concentrated positioning: bullish crypto, but on “less than five names.” He keeps Tether (“continues to be a really good business,” a regulatory arb for remittances) and Hyperliquid. Across the discussion, the stablecoin use cases described are gambling/market access, capital that cannot offramp, and people wanting faster dollar payouts; Santi doubts stablecoin transfers themselves make much money because they are “pretty much at cost.”
• Yanowitz’s rebuttal is a live case study: a LatAm AP/bill-pay company (“Ramp plus Bill.com in one”) growing 40% month-over-month at $80–90M run-rate revenue, built entirely on stablecoins its customers don’t know exist. Rain — “the biggest investment we’ve ever made as a firm” — is adding roughly 20% monthly growth; Visa-settled stablecoin volume is “doubling every few months”; and Airwallex’s Jack Zhang flipped from anti-stablecoin to musing about his own layer one. The frame: stablecoins aren’t peer-to-peer payments, they’re “a brand-new settlement layer.” The transcript also mentions an unidentified global-marketplace deal without naming the counterparty.
• On Robinhood’s AI agent and Liquid’s “co-invest” (Hyperliquid perps traded inside ChatGPT and Claude via claimed CFTC no-action relief), Yanowitz’s verdict on the legal theory is blunt: “good luck.” Santi splits the question in two — natural-language retail trading through AI front ends won’t dominate (“people will still want their finance to exist in finance”), but agentic/systematic trading on exchanges “is inevitable” — citing a crypto exchange CEO who trained an OpenClaw agent on historical data and “was losing a bunch of money.” Yanowitz won’t hand his money to chat apps: “church and state.”
• Morpho’s “Midnight” white paper revives fixed-rate, fixed-term on-chain credit — a graveyard category that includes Notional, Element Finance, and Yield attempts. Santi says the historical problems were no universal benchmark rate and extremely short-duration on-chain capital. Sky/Maker governance risk is the tell — benchmark moves equivalent to “the Fed hiking rates from 5% to 15%” in months — and when Yanowitz asked mortgage originator Better whether its Sky facility had a rate cap, “he didn’t tell me.” The mechanism itself, credit units versus debt units at a single maturity: “this is Pendle.”
• The DeFi-safety scare is real but misdiagnosed, both agree. Against the ex-OpenZeppelin founder’s “nothing in DeFi is safe” (he advises zero capital on-chain), Yanowitz argues almost no major hacks are smart-contract exploits — it’s OPSEC and social engineering, now AI-accelerated, so hacks rise across all software before “eventually hacks will basically go away altogether.” Santi’s practical read: on-chain irreversibility makes the transition brutal, so “wait it out for six months” — and the speakers recommend endpoint monitoring and stronger device and identity security, since Drift’s attackers lurked before striking.
• From D.C., Rob now thinks Polymarket’s ~55% odds on the CLARITY Act are underpriced — a reversal from when he called 50–55% too high. Floor vote likely July (the industry’s June call is “too quick”), with the ABA lobbying against it and Democratic ethics language on profiting from crypto the two live risks; on the regulatory side he expects constructive CFTC perps guidance, Polymarket-style onshore/offshore structures, and medium-term perps on real-world assets plus 24/7 spot.
• Saylor pivoting to “this week we’re not buying Bitcoin, we’re buying bonds” shook the faithful — “it’s like Jesus saying you can’t walk on water.” Strategy’s U.S.-dollar dividend coverage is down to six months, so more BTC sales “seems likely”; its NAV is cited at 1.2. Yanowitz’s standing rule: “all financial engineering… eventually blows up or unwinds — if you want exposure to Bitcoin, just buy the goddamn thing, buy IBIT at least.” The robotics vehicle RoboStrategy trades near 4× NAV, possibly higher after 25% advisory-team dilution; Santi and Yanowitz favor a synthetic tracking private-company marks instead.

Deep dive

1. A better bear market than 2018 — unless crypto’s fate is irrelevance

• Santi’s baseline: 2018 was existential — “I remember sitting with Mike getting beers… I said I would go get a job at Uber and Mike said he would go get a job at Netflix.” Today’s question is narrower: “what will we do if the tokens don’t go up again?” The thesis that all capital markets — stocks, bonds, currencies, commodities — move on-chain “has been validated more than ever in the last two years”; the open issue is whether value accrues to tokens or reverts to equity.
• Yanowitz agrees that the industry is in “a far, far better place than we’ve ever been,” but names the source of timeline bearishness: the people from 2018 aren’t exposed to what’s actually working. “A lot of people in the industry feel like they’re getting left behind” while payments and tokenized assets keep growing.
• Santi’s dissent frames the episode: “what’s worse than being dead is being irrelevant.” His team has assessed ~30 sectors expecting stablecoins to help and repeatedly found “we just don’t feel like stablecoins serve a purpose” — not a technology problem, a stakeholder-adoption one. The stablecoin buckets described in the exchange are capital stuck on-chain because it can’t offramp and people who would rather be paid faster or in dollars; gambling is the especially powerful crypto use case, alongside market access for people in certain parts of the world. “Everything else—” (Yanowitz cuts him off.)
• Yanowitz’s analogy for why gambling ripped hardest: the internet let information flow, so porn ripped first; “what rips the hardest when crypto lets value flow? Gambling.”

2. Yanowitz’s stablecoin evidence file vs. Santi’s five-name concentration

• Yanowitz’s counter is that the doubters aren’t in the rooms where it’s happening — a Brookings Institution researcher he sparred with on Twitter “didn’t even know that stablecoins were being settled directly with Visa.” His load-bearing example: a company doing AP/bill-pay for large LatAm corporates — “think Ramp plus Bill.com in one” — growing 40% month-over-month at $80–90M run-rate revenue, with the entire infrastructure built on stablecoins. “Their customers do not know.”
• The momentum list as Yanowitz tells it: Rain is adding roughly 20% monthly growth; the transcript mentions a deal with an unidentified global marketplace; Visa-settled stablecoin volume is “doubling every few months”; and Airwallex’s Jack Zhang — “the biggest anti-stablecoin person on Twitter for a little bit” — is now tweeting about building his own layer one. The reframe: stop thinking peer-to-peer transfers; stablecoins are “enabling a brand-new settlement layer.”
• Santi concedes usefulness but insists on sizing: “I’m not by any way suggesting that stablecoins are not useful… the nuance is how big and how concentrated.” He stays “very bullish but much more concentrated — probably less than five names in crypto,” with Tether — a regulatory arb, especially for remittances — and Hyperliquid as examples. His skeptic’s question on transfers: “you’re not making that much money with stablecoin transfers” — “we talked about Tempo… not going to make that much money just [on] stablecoin transfers.”
• Yanowitz’s answer is the fintech-wedge pattern: Stripe started as a gateway and became credit, accounting, virtual accounts, cards; the LatAm company already upsells expense-management software. “If you take these wedge products of just having better transfers and better settlement, you can build really big businesses.” Santi’s end-state call is that global banks will serve SMBs and retail consumers far better than existing banks, because of stablecoins.

3. The concession both sides make: hyperfinancialized capital markets are the TAM

• Yanowitz’s admission of what 2018 got wrong: not that crypto-fication of everything failed, but that “global capital markets are so much bigger than anyone realizes — orders of magnitude bigger — that if all we do is move global capital markets onto public blockchains, it’s an insane opportunity.”
• Santi agrees via the earlier hyperfinancialization take: “anyone with an internet connection can go long a prediction market on SpaceX pre-IPO” — the TAM keeps growing even if the winners stay concentrated. Yanowitz adds the pipeline: Ondo launching tokenized RWAs, synthetics, and markets with OKX, plus expected CFTC perps guidance covering tokenized synthetics settling on-chain — and those markets “will also need to be settled in stablecoins.”
• The full-circle diagnosis from Yanowitz: the bearish timeline exists because the winning thesis “is boring… and doesn’t necessarily accrue value to the tokens that people want to see go up.”

4. Robinhood’s agent, Liquid’s co-invest, and whether Claude eats fintech

• The news: Liquid’s “co-invest” puts Hyperliquid perps trading directly inside ChatGPT and Claude, claimed legal in all 50 states via CFTC no-action relief; Robinhood announced an AI-agent setup that can accept an outside agent, give it a siloed wallet or account, and impose spending limits. On the no-action theory, Yanowitz offers only: “good luck.” He later adds, “Godspeed.”
• Santi’s big question: social media gobbled up media companies nobody predicted; could ChatGPT and Claude gobble the fintechs — “will we have SoFi in 10 years? Will we have Robinhood in 10 years?” Yanowitz’s own answer is no, personally: “I don’t want to trade through Twitter… church and state” — though “I could also see myself being completely wrong here.”
• Santi’s split: natural-language trading through AI front ends probably won’t dominate — existing startups are “pretty terrible” — but ML-powered agentic trading migrating from institutional to retail “is inevitable.” His anecdote: a large crypto exchange CEO trained an OpenClaw agent on historical trading data and gave it his own accounts — “to his credit, he told me they were losing a bunch of money.”
• Yanowitz sees a business in guardrails — risk-profile parameters like brokerage onboarding — but with a durability caveat: Anthropic and OpenAI may crush whatever gets traction. He also raises the long-arc analogy of Bloomberg eventually realizing, “Why don’t we just let people trade through the platform?”

5. Morpho’s fixed-rate push into a graveyard category

• Yanowitz’s setup: fixed rates are DeFi’s “missing primitive.” On Compound or Morpho Blue, rates track pool utilization, so borrowers carry interest-rate risk and businesses simply don’t borrow. The trad analog is the 30-year mortgage — described here as floating rates fixed via swaps — and the unlock now is protocols with real revenue: a Uniswap making a hypothetical $100M should be able to plan around a fixed loan, opening business borrowing, mortgages, and institutional/RWA lending.
• Santi says many prior attempts “really worked” poorly. The transcript’s other speaker says the firm invested in Notional and Element Finance; Santi recalls that Yield tried this. The live post-mortem identifies no universal benchmark rate — attempts used ETH staking yield, Maker, or Compound as ersatz central banks — and on-chain capital markets that are “extremely short duration.” Yield-farm loops don’t need 30-year fixed rates. Governance risk compounds it: Maker/Sky rate changes were “equivalent [to] the Fed hiking rates from 5% to 15%” in a couple of months.
• Yanowitz’s live example: mortgage originator Better tapped Sky for a sizable facility, and when he asked whether the deal guaranteed a rate cap — since Sky governance could swing 3% to 10% “on a whim” — “he didn’t tell me.” Rob: “that tells you.”
• On the mechanism — a single maturity date, buying credit units and selling debt units, redeemable 1:1 at term — the consensus is instant: “this is Pendle.” Yanowitz’s caution from Pendle’s history: it grew on Ethena yield-farming exposure, so the open question is whether “regular-way economic activity” comes on-chain or whether it stays trading and farming.

6. “Nothing in DeFi is safe” — right fear, wrong diagnosis

• The trigger: the OpenZeppelin founder — whom Yanowitz notes left in 2019 — tweeted that nothing in DeFi is safe and that he had advised family, friends, and himself “to put zero capital on-chain” — full stop, not “at the moment,” Santi observes. Santi says he has privately heard credit funds and other historically active DeFi participants focusing intensely on questions about stablecoins on Aave and whether the Kelp situation is resolved. Kelp is not fully resolved legally, though users were essentially made whole.
• Yanowitz’s rebuttal: “most of the hacks, almost all of them, are not smart-contract hacks” — they’re OPSEC failures involving a compromised user, key holder, or signer. AI cuts both ways: attackers “only have to find one vulnerability” and, once inside a security perimeter, AI lets them find and extract capital “faster than ever.” His arc: more hacks across all software for a period, “and then eventually hacks will basically go away altogether — it just takes time to rewrite code.”
• Santi’s practical overlay: on-chain irreversibility makes the transition uniquely dangerous, so “wait it out for six months.” He cites a CISO’s advice to use endpoint security and monitoring because an attacker may already be lurking; Drift is given as an example of a delayed strike. The other speaker adds separate devices when using Claude Code, continuous monitoring, and a safe word to prove humanity; vibe-coding widens the attack surface further.

7. D.C. read: CLARITY now underpriced at ~55%, perps guidance coming

• Rob, back from meetings with Hill staff and regulators, has flipped his own call: “in the past when Polymarket showed that CLARITY was 50–55%… I was saying, oh, that’s too high. I now think the likelihood is higher than 55” — the market, from a low around 36% to roughly 56%, is “underpricing it a little bit.” Floor vote expectation: July, not the industry’s June call (“too quick”).
• The two live fights: the ABA — bankers still lobbying hard even after Tillis moved the bill out of the Banking Committee — and Democratic ethics language concerning the ability to profit from crypto and crypto businesses, which the White House will have to accept. Floor debate will also hit BRCA developer protections for open-source code, validators, front ends, and smart contracts; Rob calls the language “solvable.”
• On regulation, Yanowitz says CFTC Chair Selig discussed perps guidance in March. He expects it to be constructive — onshore/offshore entity structures “the way Polymarket does today” — with perps and prediction markets dominating near-term, then “perps for real-world assets” and 24/7 spot in the medium-to-long term. His closing tension: “a lot of positive things are happening, and yet the price action hasn’t made people as excited as you’d think.” Santi’s cold comfort: the Nasdaq took 15 years to reclaim its highs.

8. Saylor buys bonds; the DAT-unwind rule; a robotics vehicle at 4× NAV

• The week’s shock: Saylor’s “this week, we’re not buying Bitcoin, we’re buying bonds,” plus a billion-dollar IBIT trade circulating. Santi says it rattled people who had made Saylor “basically Bitcoin Jesus”: “it’s like Jesus saying you can’t walk on water.” The dashboard detail: Strategy cut its U.S.-dollar dividend coverage to six months — more Bitcoin sales “seems likely” — while its NAV is cited at 1.2.
• Yanowitz’s standing rule: “all financial engineering, whether it’s a DAT or anything, eventually blows up or unwinds — and usually doesn’t unwind easily… if you want exposure to Bitcoin, just buy the goddamn thing, buy IBIT at least.”
• The same lens applies to RoboStrategy, the robotics vehicle trading close to 4× NAV — perhaps higher after accounting for 25% dilution to the advisory team. Santi says there is no compelling argument for that premium and revives Yanowitz’s earlier synthetic idea: someone could create a synthetic on Apptronik, Figure, or another private company that tracks the latest recorded trade. “Crypto is really good at finding liquidity for a tail end of assets,” making a synthetic potentially better than paying a venture-access premium.